Mood stabilizer treatment is central to the pharmacological management of patients with bipolar disorder. However, response to such agents is highly variable between individuals often resulting in a lengthy trial and error process of medication optimization that can last years. There is a great need for a better predictor of response which would guide physicians to the optimum medication in a more efficient fashion. Genetic differences likely explain a substantial portion of this variability. The goal of this project is to identify genetic variants that are associated with response to mood stabilizer medications that might ultimately be useful as a predictive test. Studies to date by our group have implicated two genes, NTRK2 and PDE11A as predicting response to lithium. In this project, we propose a two pronged approach: genes will first be sought in an exploratory fashion in a larger retrospective sample and then tested for replication in a smaller prospective sample. Larger samples are more easily obtainable in a retrospective study, however, prospective designs though more difficult, provide better and more quantitative data. Our 11 site consortium has recently completed collection of over 4,500 bipolar subjects for a large genetic study. 2,000 retrospective subjects will be collected from both recontact of these previous cases and recruitment of new retrospective cases. The prospective sample of 960 subjects will be collected through an eight site multicenter trial. Patients will be recruited, screened and stabilized first on lithium monotherapy over a 3 month period. After one month observation, they will enter the maintenance phase and followed for 2 years. The primary outcome measure will be time to relapse. Patients who fail lithium will be crossed over to valproic acid, those failing both drugs will enter the treatment as usual arm. Genomewide association will be performed on the retrospective sample and positive SNPs replicated in the prospective sample. Secondary analyses will include genomewide association ofthe prospective sample alone and in meta-analysis with the retrospective sample. These analyses will be guided in part by studies of lithium's mechanism of action in neuronal cells derived from induced pluripotent stem cells in turn derived from skin fibroblasts from lithium responders and non-responders. RELEVANCE (See instructions): This multi-site collaborative project aims to identify genetic variants in individuals with bipolar disorder that predict response to lithium. We will do this with a combination of retrospective assessment of lithium response in 1600 individuals with BP disorder and analysis of genotype data, as well as a prospective study of 1000 BP individuals who begin an open trial with lithium. Our hypothesis is that genetic variants at several loci predict treatment outcomes with lithium.